Google is reportedly in talks with the United States Department of Defence to use its advanced Gemini AI models in classified environments, which means for secure and confidential military operations. According to The Information report, the agreement is under discussion, and if the deal is finalised, the Pentagon could utilise Google’s AI in legal and regulatory frameworks.

It suggests that Google has proposed specific rules in the contract that will limit how its AI model can be used, to avoid controversial use. The suggested rules will limit the use of AI for domestic mass surveillance or for autonomous weapons without proper human oversight. 

Also read: Banned but booming: Apple, Google still show ‘nudify’ apps in search results

The Information quoted a Pentagon official stating that it will continue using advanced AI technologies through partnerships with companies, but they did not confirm whether they are actually in talks with Google. However, it will be the tech giant to expand its partnership with the US government.

In February 2026, OpenAI locked a “All Lawful Purposes” deal with the Pentagon to deploy its AI models for classified operations. It was suggested that the deal would allow the DoD to use AI for mass surveillance, and many of OpenAI's users started to uninstall its tools and services.

OpenAI also provided clarity on its deal that it has banned the use of its AI for domestic mass surveillance and autonomous weapons. And that  AI will be used only through cloud systems, not deployed freely or independently.

Also read: Anthropic signs multi-gigawatt AI compute deal with Google, Broadcom

On the other hand, Anthropic’s engagement with the Pentagon took a more guarded turn, as the company refused to ease its safeguards. As a result, it is now facing a potential ban from the US government, with the Pentagon labelling it a “supply-chain risk.” Anthropic has since taken legal action against the government, alleging the move was unlawful retaliation for its public safety stance.